WebCanvas: Benchmarking Web Agents in Online Environments

ACL ARR 2024 December Submission1862 Authors

Abstract: For web agents to be practically useful, they must adapt to the continuously evolving web environment characterized by frequent updates to user interfaces and content. However, most existing benchmarks only capture the static aspects of the web. To bridge this gap, we introduce WebCanvas, an innovative online evaluation framework for web agents that effectively addresses the dynamic nature of web interactions. WebCanvas contains three main components to facilitate realistic assessments: (1) A novel evaluation metric which reliably capture critical intermediate actions or states necessary for task completions while disregarding noise caused by insignificant events or changed web-elements. (2) A benchmark dataset called Mind2Web-Live, a refined version of original Mind2Web static dataset containing 542 tasks with 2439 intermediate evaluation states; (3) Lightweight and generalizable annotation tools and maintenance pipelines that allow the community to collect and maintain the high-quality, up-to-date datasets. Building on WebCanvas, we open-source a baseline agent framework with extensible modules for reasoning, providing a foundation for the community to conduct online inference and evaluations. Our best performing agent achieves a task success rate of 22.1% and a task completion rate of 50.0% on the Mind2Web-Live test set. In addition, we analyze performance discrepancies in various websites, domains, and experimental environments. We encourage the community to contribute further insights on online agent evaluation, thereby advancing this field of research.

Paper Type: Long

Research Area: Resources and Evaluation

Research Area Keywords: evaluation methodologies, benchmarking, NLP datasets, metrics, embodied agents

Contribution Types: NLP engineering experiment, Publicly available software and/or pre-trained models, Data resources

Languages Studied: English

Submission Number: 1862